Gartner analyst criticizes AI inefficiency and agentic hype

"AI is not doing its job today and should leave us alone", declared Erick Brethenoux, Gartner's global chief of AI research, during the firm's Data & Analytics Summit in Sydney.

Key Criticisms of Current AI

• Brethenoux criticized AI-generated meeting summaries for merely listing actions rather than executing tasks: "Just go and do it already"
• Highlighted successful "Empathy AI" implementations at:
  • Vizient (US healthcare) - automated most-dreaded employee tasks
  • A real estate firm - parallelized 17-step tenant assessment process

Agentic AI Challenges

The analyst delivered sharp warnings about enterprise AI agents:

• "Industrial companies have used them for decades... seldom found the software can handle very complex tasks"
• Raised critical unanswered questions:
  • How to orchestrate 50,000 agents across an enterprise
  • How agents negotiate competing priorities (work vs personal)
  • Vendors respond with "silence" to these challenges

Fundamental Issues

Brethenoux emphasized this is primarily a software engineering problem requiring:

• Clear system decomposition
• Defined communication protocols
• Precise autonomy levels
• Controlled perception/execution parameters

"It's not trivial", he stressed, while accusing vendors of promoting unrealistic agentic nirvana.

Related Coverage

• BT chief says AI could deliver more job cuts
• Half of businesses rethink ditching humans for customer service bots
• Go ahead and ignore Patch Tuesday

Closing with an Albert Camus quote: "To misname things is to contribute to the world's miseries", Brethenoux condemned the industry's fuzzy definitions of "AI agent" and "generative AI".